OPAG shaped a governed AI referral follow-up agent for Retina Eyecare that prepared 33 closed-loop outreach packets across aging referrals, missing imaging, incomplete patient contact, scheduling readiness, insurance context, provider review, and audit status. The agent supported care coordination; it did not diagnose, make clinical decisions, or contact patients outside approved review rules.
Key takeaways
- The case study focuses on one feature: closed-loop referral follow-up, not general clinic automation or autonomous clinical triage.
- The agent connected OPAG Conversational AI for source-linked referral questions with Agentic AI for outreach packet routing, provider review, escalation rules, and audit trails.
- This case study links to OPAG guidance on referral leakage monitoring AI, healthcare AI intake, and the related Retina chart-prep case study because referral follow-up needs patient access, imaging, scheduling, and provider accountability in one controlled loop.
What did the OPAG referral follow-up agent do for Retina Eyecare?
Specialty clinics can lose momentum after a referral is received. A patient may be missing imaging, a referring provider may need a status update, insurance context may be incomplete, or staff may not know whether outreach is ready.
OPAG narrowed the workflow to one agent capability: closed-loop referral follow-up. The agent prepared 33 outreach packets so Retina Eyecare teams could see why a referral was aging, what evidence was missing, who owned the next action, and whether provider review was required.
The answer-first summary is this: OPAG used governed AI to make referral follow-up more complete, source-linked, and auditable while keeping patient outreach and clinical decisions inside approved human review.
Why does closed-loop referral follow-up AI matter for specialty clinics?
Retina and eye-care referrals often depend on imaging, prior records, patient availability, urgency cues, insurance or authorization status, and provider review. If any part is missing, the referral can stall.
OPAG designed the workflow so the agent could identify follow-up blockers, assemble source-linked packets, and route the right next action to staff or providers before the referral slipped further.
- Patient access teams needed aging referral queues with clear next-best coordination steps.
- Imaging teams needed visibility into missing or mismatched retinal imaging context.
- Providers needed review queues for referrals with clinical sensitivity or unclear urgency.
- Operations leaders needed audit trails for outreach attempts, reviewer edits, escalations, and closure status.
How did the agent prepare 33 closed-loop outreach packets?
The workflow started with the referral lifecycle rather than a generic chatbot. OPAG defined which records were allowed, who could view patient context, which missing items mattered, and when provider review had to happen before outreach.
Each packet included a short reason for follow-up, linked source evidence, missing items, proposed next action, owner, approval requirement, and audit status. That made referral closure inspectable before staff contacted the patient or referring office.
- Scan: review referral date, source, reason, imaging references, appointment status, contact attempts, insurance notes, and provider rules.
- Classify: group packets as missing imaging, incomplete contact, ready to schedule, provider review, insurance question, or closed.
- Draft: prepare a source-linked outreach packet with missing evidence, proposed next action, owner, and approval need.
- Route: send administrative follow-up to patient access, imaging gaps to imaging staff, and sensitive cases to provider review.
- Audit: record source retrieval, proposed action, reviewer edit, outreach status, escalation, closure, and override reason.
What governance protected patient outreach and clinical context?
Closed-loop referral follow-up has to balance speed with patient-sensitive boundaries. OPAG kept the agent focused on preparation and routing while staff and providers owned outreach, clinical judgment, and care decisions.
The control layer defined what the agent could read, summarize, classify, route, and log. Patient messaging, clinical prioritization, urgent escalations, and care decisions required approved human review.
- Role-based access limited patient, imaging, insurance, and provider context to authorized users.
- Data minimization kept the agent focused on referral follow-up and closure evidence.
- Provider review gates protected clinical urgency, diagnosis-sensitive context, and care decisions.
- Approved outreach rules controlled when the agent could draft follow-up packets for staff action.
- Audit logs captured source context, reviewer edits, patient access actions, escalations, closure status, and overrides.
Which OPAG services connect to closed-loop referral AI?
The referral follow-up agent shows how OPAG turns care-coordination friction into a controlled operating workflow. Conversational AI helps staff ask source-linked questions. Agentic AI routes follow-up packets, provider reviews, and escalation items. Governance keeps patient-sensitive work inside accountable review boundaries.
The same service pattern can support hospitals, specialty clinics, diagnostic labs, care-coordination teams, prior authorization support, provider chart-prep, and post-visit follow-up workflows.
- Conversational AI: source-linked answers about referral status, missing items, imaging context, and scheduling readiness.
- Agentic AI: review queues, follow-up routing, escalation thresholds, closure tracking, and audit logs.
- Referral leakage monitoring AI: aging referrals, outreach packets, specialty routing, and measurable closure outcomes.
- AI readiness assessment: choosing a first clinic workflow with clear data, owners, controls, and ROI signals.
Frequently asked questions
Did the OPAG referral follow-up agent contact patients automatically?
No. The agent prepared follow-up packets and routed review. Patient outreach, provider escalation, clinical prioritization, and care decisions stayed under approved human review rules.
What data does a referral follow-up AI agent need?
Useful sources include referral records, imaging references, appointment status, contact history, insurance or authorization notes, consent status, provider-review rules, outreach templates, closure status, and audit history under role-based permissions.
Which OPAG capabilities power this referral follow-up case study?
The case study combines Conversational AI for source-linked referral questions, Agentic AI for governed routing, and healthcare AI intake patterns for privacy-aware preparation.
Can this closed-loop referral pattern work beyond eye-care clinics?
Yes. The same referral follow-up pattern can support hospitals, specialty clinics, diagnostic labs, care coordination, post-visit follow-up, prior authorization support, and provider-preparation workflows when review owners and data boundaries are defined.
